Enrichment of Amadori products derived from the nonenzymatic glycation of proteins using microscale...

Analytical Biochemistry 393 (2009) 8–22

Contents lists available at ScienceDirect

Analytical Biochemistry

journal homepage: www.elsevier .com/locate /yabio

Enrichment of Amadori products derived from the nonenzymatic glycationof proteins using microscale boronate affinity chromatography

Anikó Takátsy a,1, Katalin Böddi a,1, Lívia Nagy b, Géza Nagy b, Szilvia Szabó c, Lajos Markó d,István Wittmann d, Róbert Ohmacht a, Thomas Ringer e, Günther K. Bonn e, Douglas Gjerde f, Zoltán Szabó a,*

a Department of Biochemistry and Medical Chemistry, University of Pécs, 7624 Pécs, Hungaryb Department of General and Physical Chemistry, University of Pécs, 7624 Pécs, Hungaryc Department of Nutritional Sciences and Dietetics, University of Pécs, 7623 Pécs, Hungaryd Second Department of Medicine and Nephrological Center, University of Pécs, 7624 Pécs, Hungarye Institute of Analytical Chemistry and Radiochemistry, Leopold–Franzens University, 6020 Innsbruck, Austriaf PhyNexus, San Jose, CA 95136, USA

a r t i c l e i n f o

Article history:Received 10 February 2009Available online 12 June 2009

Keywords:Boronate affinity chromatographyGlycationMALDI–TOF/MSType 2 diabetes mellitusAmadori products

0003-2697/$ - see front matter � 2009 Elsevier Inc. Adoi:10.1016/j.ab.2009.06.007

* Corresponding author. Fax: +36 72 536 226.E-mail address: [email protected] (Z. Szabó

1 These authors contributed equally to this work.2 Abbreviations used: AC, Amadori compound; AGE, ad

HbA1c, hemoglobin A1c; HSA, human serum albuminMS, mass spectrometry; CE, capillary electrophoresis;desorption/ionization; RNase A, ribonuclease A; ACN, acacid; CHCA, a-cyano-4-hydroxycinnamic acid; DHB, 2sinapinic acid; DTT, D/L-dithiothreitol; ACTH, adrenobovine serum albumin; HPLC, high-performance liquiviolet; PBS, phosphate-buffered saline; SPE, solid phflight; FIA, flow injection analysis; MS/MS, tandem malysine.

a b s t r a c t

Amadori peptides were enriched using boronate affinity tips and measured by matrix-assisted laserdesorption/ionization time-of-flight mass spectrometry (MALDI–TOF/MS). As demonstrated by electro-chemical measurements, the tips show the highest binding efficiency for glucose at pH 8.2 employingammonium chloride/ammonia buffer with ionic strength of 150 mM, exceeding taurine buffer at thesame concentration. The bound constituents were released by sorbitol and formic acid. It was also dem-onstrated that elution with sorbitol at 1.2 M is superior to acidic media. Comparison of results was basedon the numbers of detected peptides and their glycated sites. Using sorbitol for elution requires desaltingprior to analysis. Therefore, three different sorbents were tested: fullerene-derivatized silica, ZipTip(C18), and C18 silica. Fullerene-derivatized silica and ZipTip showed the same performance regardingthe numbers of glycated peptides, and sites were better than C18 silica. The elaborated off-line methodwas compared with liquid chromatography–tandem mass spectrometry (LC–MS/MS) measurements, bywhich considerable less modified peptides were detected. Affinity tips used under optimized conditionswere tested for the analysis of human serum albumin (HSA) from sera of healthy and diabetic individuals.A peptide with a mass of 1783.9 Da could be detected only in samples of diabetic patients and, therefore,could be a very interesting biomarker candidate.

� 2009 Elsevier Inc. All rights reserved.

Nonenzymatic glycation is the covalent binding of single reduc- more reactive constituents termed advanced glycation endproducts
ing sugars (e.g., glucose, fructose, ribose) to primary amino groupsin proteins. The initial product of a nonenzymatic glycation reac-tion is a labile Schiff base intermediate that slowly isomerises,and therefore forms, a stable ketoamine called the Amadori com-pound (AC)2 [1]. The AC can undergo additional oxidation and rear-rangement reactions to form a series of biologically considerably
ll rights reserved.

).

vanced glycation endproduct;; LC, liquid chromatography;MALDI, matrix-assisted laseretonitrile; TFA, trifluoroacetic

,5-dihydroxybenzoic acid; SA,corticotropic hormone; BSA,

d chromatography; UV, ultra-ase extraction; TOF, time-of-ss spectrometry; FL, fructosyl

(AGEs) [2]. They are known to be responsible for the developmentof diabetic complications such as diabetic nephropathy and retinop-athy [3]. Since the 1970s, the measurement of glycated hemoglobinA1c (HbA1c) has been used routinely as a clinical diagnostic markerfor relatively long-term (4–6 weeks) glucose control in diabetic pa-tients [4]. However, to find more specific and informative proteinbiomarkers for monitoring the glycemic state and to get deeper in-sight into the role of glycation in the development of diabetic com-plications, comprehensive proteomic studies are required foridentifying those glycated proteins whose altered structures maycontribute to pathology.

Glycated human serum albumin (HSA) is an important shortertime indicator of diabetes that is more sensitive to changes inblood glucose level than HbA1c [5]. The glycated albumin level alsoprovides useful information on glycemic control when monitoringthe efficacy of therapy [6]. Recently, the main techniques investi-gating the glycation of proteins are based on a variety of on- andoff-line mass spectrometric methods (e.g., liquid chromatography[LC] coupled to mass spectrometry [MS], capillary electrophoresis

http://dx.doi.org/10.1016/j.ab.2009.06.007

mailto:[email protected]

http://www.sciencedirect.com/science/journal/00032697

http://www.elsevier.com/locate/yabio

Enrichment of Amadori products / A. Takátsy et al. / Anal. Biochem. 393 (2009) 8–22 9

[CE], LC separation/fractionation–matrix-assisted laser desorption/ionization [MALDI] MS) [7–10]. Although a great number of studiesdeals with the investigation of glycation as well as with methodicaldevelopment for the investigation of glycated constituents, there isa need to improve the existing methods or to develop novel, moresufficient, and less time-consuming techniques.

Boric acid and boronic acid are well known to form stable esterswith polyols and saccharides containing a cis–diol part [11]. Affin-ity attachment (equilibria) between cis–diol-containing peptideand boronic acid and the mechanism of the elution are shown inScheme 1. Increased attention has been paid to these compoundsbecause of their prospective applicability to recognize sugars[12]. Although m-aminophenylboronic acid is immobilized mainlyon agarose beads [13] exceptions include carboxylic acid-termi-nated magnetic beads [14] and polymer monolith [15]. In a previ-ous attempt, hemoglobin from diabetic and nondiabeticindividuals was separated by affinity chromatography employingimmobilized phenylboronate [16]. In this work, the effect of pHand ligand concentration and the influence of the concentrationof Mg2+ ions of the binding buffer on the amount of bound glycatedhemoglobin was discussed in detail, but the conditions of the elu-tion received less attention. Selective enrichment of glycated pep-tides was done with an affinity LC column coupled directly to anelectron transfer dissociation mass spectrometer [17]. Anotherinteresting approach was the combination of boronic acid and lec-tin in a single column to isolate glycoproteins by either selective or

Scheme 1. Mechanism of binding and elution processes betwe

combined elution modes [18]. The application of the immobilizedphenylboronic acid on agarose and sepharose hydrogel microparti-cles, which are deposited on an aluminum chip, enabled the high-throughput analysis of AGEs from serum sample by MALDI–MS.One of the most advantageous features of the microchip men-tioned above is that no elution of the bound molecules is required[19].

The binding ability of the immobilized boronate ligand towardcis–diol-containing biomolecules such as catechols, nucleotides,glycoproteins, and some enzymes is pH dependent [20]. In general,a boronate phase can bind molecules under slightly alkaline pH.Unfortunately, compounds such as L-DOPA-containing peptidesare rather unstable under alkaline conditions. This difficulty couldbe overcome by introducing electron-withdrawing groups into thephenyl ring of phenylboronic acid. In this work, the solid support towhich the boronic acid had been attached was a porous silica gel[21].

Shielding boronate affinity chromatography is a useful tool thatcan suppress nonspecific interactions between the boronate ligandand nonglycosylated proteins. This technique is based on the intro-duction of a so-called shielding reagent, a low-molecular-masspolyhydroxyl compound that can interact with the boronate ligand[22]. The shielding reagent avoids the nonspecific interactions be-tween the boronate ligand and nonglycosylated proteins.

While using boronate affinity chromatography for glycopro-teins, secondary interactions (especially hydrophobic and ionic

en cis–diol-containing Amadori peptides and boronic acid.

10 Enrichment of Amadori products / A. Takátsy et al. / Anal. Biochem. 393 (2009) 8–22

interactions) must be considered. Hydrophobic interactions aresometimes responsible for the nonspecific binding of undesiredproteins, and this phenomenon can be avoided entirely by addingdetergents to the mobile phase/binding buffer [13,23]. Ionic inter-actions between the negatively charged boronate and anionic pro-teins can hinder the necessary binding, and this is why Mg2+ ions atlow concentrations are to be added to the mobile phase/bindingbuffer to get rid of the undesirable ionic effects [13,23].

The current study gives a comprehensive description of thecharacterization and application of microscale boronate affinitychromatography fabricated in a pipette-like tip. In this study, thebinding capacities of these tips are measured electrochemically;therefore, the most important experimental conditions, such as io-nic strength, pH of the binding buffer, and elutions with formicacid and sorbitol, were optimized. Two different buffer systemswere used for the binding of glycated peptides. The method dem-onstrated first with standard in vitro glycated peptides receivedfrom in vitro glycated digests of ribonuclease A (RNase A) andHSA were compared using the LC–MS method. Finally, the applica-bility of the boronate affinity tips was introduced for the investiga-tion of glycated HSA isolated from the serum of nondiabetic anddiabetic patients.

Materials and methods

Chemicals

Acetonitrile (ACN), methanol (both gradient grades), HSA (97–99%), RNase A from bovine pancreas (>90%), trifluoroacetic acid(TFA, P99%), a-cyano-4-hydroxycinnamic acid (CHCA, P99%),2,5-dihydroxybenzoic acid (DHB, P99%), sinapinic acid (SA,P99%), D/L-dithiothreitol (DTT, >99%), taurine (P98%), iodoaceta-mide (P98%), 2-mercaptoethanol (P98%), and ammonium bicar-bonate (>99%) were purchased from Sigma–Aldrich (Budapest,Hungary). D-Glucose monohydrate (puriss), D-sorbitol (puriss)urea, sodium hydroxide (analytical grade), magnesium sulfate(analytical grade), ammonium chloride (analytical grade), ammo-nium hydroxide (25%, analytical grade), toluene (>99%), and tetra-hydrofuran (>99.8%) were obtained from Reanal Finechemical(Budapest, Hungary). Formic acid (98–100%) was bought fromScharlau Chemie (Barcelona, Spain). Peptide calibration standard(consisting of bradykinin, angiotensin II, angiotensin I, substanceP, renin substrate, adrenocorticotropic hormone [ACTH] clip (1–17), ACTH clip (18–39), and somatostatin) and protein calibrationstandard (consisting of trypsinogen, protein A, bovine serum albu-min [BSA], and BSA–dimer) were obtained from Bruker Daltonics(Bremen, Germany). Trypsin (sequencing grade, modified) wasprovided by Promega (Madison, WI, USA). Double distilled waterwas prepared in our laboratory.

Instrumentation

An AUTOLAB12 electrochemical workstation controlled withGPES software (version 4.9.009 for Windows, Eco Chemie, Utrecht,Netherlands) served as an amperometric measurement instru-ment. A homemade flow injection analysis (FIA) manifold [24]was used for amperometric glucose analysis of eluted samples. Awall-jet-type detector cell containing working, counter, and refer-ence electrodes was applied in the manifold. A platinum wire of1 mm diameter served as counter electrode, as did a silver wirefor semireference. A copper microdisk working electrode of30 lm diameter and 0.1 M sodium hydroxide carrier solutionwas employed in the FIA glucose measurements.

For analyzing glucose samples obtained with sorbitol elutionfrom the boronate affinity tips, amperometric glucose biosensor

was used. The reason why a selective biosensor needed to be em-ployed is that sorbitol also reacts on copper electrodes in basicmedia; therefore, its oxidation current would have interfered inthe case of FIA measurements. The glucose electrode was preparedin our laboratory by immobilizing glucose oxidase enzyme on aplatinum disk surface (diameter) of 1 mm [25]. The measurementswith the biosensor were carried out in intensively stirred phos-phate buffer (pH 7.4). Buffer (5 ml) was pipetted into a small bea-ker used as measurement cell, small doses of calibrating standardsor samples were added, and the amperometric current at 0.65 V(vs. Ag/AgCl reference) was taken for calibration or concentrationevaluation.

An Autoflex II MALDI instrument from Bruker Daltonics wasemployed for the mass spectrometric measurements.

The high-performance liquid chromatography (HPLC) instru-ment used for the isolation and purification of HSA from serumsamples consists of a Dionex P680 gradient pump, a Rheodyne8125 injection valve, and a Dionex UVD 170U ultraviolet (UV)–vis-ible detector (Germering, Germany). Data acquisition was carriedout with Chromeleon software (version 6.60 SP3 Build 1485).

The l-LC device consisted of an Ultimate l-HPLC pump withcolumn oven, a Switchos l-column-switching device with loadingpump and two 10-port valves, and a FAMOS l-autosampler (LCPackings, Amsterdam, Netherlands). Hyphenation to the massspectrometer was carried out by a nanoflow electrospray ioniza-tion source from Proxeon (Odense, Denmark) with Pico Tips fromNew Objective (FS360-20-10, Woburn, MA, USA). Mass spectro-metric data were obtained on the linear ion trap LTQ mass spec-trometer from Thermo Fisher Scientific (Waltham, MA, USA). Adatabase search was carried out with BioworksBrowser 3.3.1 SP1(Thermo Fisher Scientific) and Sequest against the Swiss–Protdatabase.

In vitro glycation of RNase A and HSA

HSA and RNase A (5 mg) were dissolved in 1 ml of phosphate-buffered saline (PBS) buffer (pH 7.4) containing D-glucose at0.167 M [26]. Solutions of RNase A were incubated for 14 days,whereas solutions of HSA were incubated for 12 and 28 days. Incu-bations were carried out under aseptic conditions at 37 �C. Beforefurther modifications and tryptic digestion, the glycated RNase Aand HSA were purified by centrifugation through a membrane(cutoff size = 3000 Da, Millipore, Bedford, MA, USA).

Optimization of composition of binding buffer and circumstances ofelution using amperometric method

In the first experiment, an ammonium chloride solution at aconcentration of 250 mM was prepared and the pH values of thissolution were adjusted by adding ammonia to obtain binding buf-fers with pH values of 7.4, 7.8, 8.2, 8.6, 9.0, and 9.4. Each solutioncontained 50 mM magnesium sulfate. After rehydration boronateaffinity tips were equilibrated with the binding buffers mentionedabove, 0.01 mol D-glucose was dissolved in 10 ml of binding bufferand the solution was aspirated through the affinity gel. This wasfollowed by a washing step using binding buffer to eliminate thenonbound glucose. Finally, the elution of the bound D-glucosewas carried out using 200 ll of formic acid solution having a pHvalue of 2.0. In the forthcoming experiment, the pH was kept at aconstant value and the concentration of the ammonium chloridewas changed from 50 to 300 mM. Then 0.01 mol D-glucose wasagain dissolved in solution and aspirated through the gel slab ofthe tips. Elution of the bound glucose was implemented using200 ll of formic acid solution at pH 2.0. From these experiments,the binding buffer in which the tip possessed the highest bindingcapacity toward glucose was chosen and the elution was investi-


gated using formic acid from pH 2.0 to pH 5.0. Additional experi-ments for the optimization of the elution include the use of sorbitolat higher concentrations to remove D-glucose from the boronateaffinity tips in the range from 0.1 to 1.6 M. It is important to men-tion that the presence of sorbitol had no disturbing effect on theamperometric measurement carried out with the glucose biosen-sor. Water from the eluates was evaporated under vacuum, andthe remaining solid was taken up in 100 ll of double distilledwater and analyzed by means of the amperometric method usingeither a wall-jet-type detector or a biosensor.

Isolation and purification of HSA from serum samples obtained frompatients suffering from type 2 diabetes mellitus and from healthyvolunteers

Human sera taken from diabetic patients and healthy volun-teers were diluted 20 times with double distilled water, and then2 ml of this solution was centrifuged through a Centricon UltracelYM-50 centrifugal filter tube (Millipore) at 5000g for 20 min (cut-off = 50,000 Da). This centrifugation ensured the removal of serumconstituents with masses less than 50,000 Da; therefore, the life-time of the RP column used in the forthcoming step was prolonged.Proteins remaining on the filter were further diluted with doubledistilled water 20 times, and the HSA was separated on a KovasilMS-C18 nonporous column (Zeochem AG, Uetikon, Switzerland).Eluent A consisted of 5% (v/v) ACN in water and 0.1% TFA, and elu-ent B consisted of 95% (v/v) ACN and 0.1% TFA. The gradient appliedwas as follows: 0–20 min, 0% B ? 60% B; 20–25 min, 60% B ? 100%B. The flow rate was 0.7 ml min�1. Chromatograms were acquiredat 214 nm. Fractions collected from sera were evaporated to dry-ness before digestion.

Tryptic digestion of glycated HSA and RNase A

HSA or RNase A (1 mg) was dissolved in 500 ll of denaturingbuffer consisting of 8 M urea and 0.5 M ammonium bicarbonateand was shaken for 30 min at 37 �C to denature the proteins. Then50 ll of 30 mM DTT solution was added to break the disulfidebonds (37 �C, 4 h). After the solution was cooled down, 25 ll of100 mM iodoacetamide solution was added for 15 min in darknessto alkylate the cysteines. The reaction was stopped by the additionof 50 ll of 100 mM 2-mercaptoethanol, and the mixture was keptat ambient temperature for 15 min. Then the solutions of both pro-teins were put into a Microcon-YM3 centrifugal filter tube (Milli-pore) and centrifuged at 13,000g for 4 h. This ensured theremoval of excess substances through the membrane (cut-off = 3000 Da). The residues of the modified proteins were digestedwith a trypsin-to-protein ratio of 1:100 in 50 mM ammoniumbicarbonate solution overnight at 37 �C. The digestion was stoppedby evaporation under vacuum, and the resulting digest was redis-solved in 5 ll of water. In the case of HSA isolated from sera,approximately 300 lg of proteins was in-solution digested usingthe protocol described above.

Enrichment of glycated peptides using boronate affinity tips

Pipette tips containing 5 ll of gel of immobilized m-amin-ophenylboronic acid with a total volume of 200 ll (PhyTip 1000+columns, PhyNexus, San Jose, CA, USA) were used for the enrich-ment of the glycated peptides. After rehydration, the tips wereequilibrated with binding buffers. The digest of 100 lg of protein(for each protein) was dissolved in the binding buffer, and the solu-tion was aspirated through the affinity gel. Then the tips werewashed with 1 ml of binding buffer to remove unselectively boundpeptides. This was followed by an additional washing step withdouble distilled water to eliminate the trace of salts presented in

the buffer if the elution was carried out using formic acid. Glycatedpeptides were eluted with 4� 50 ll of formic acid solution at pH2.0. Elution of the bound glycated peptides has also been accom-plished using sorbitol at higher concentrations. In the case of elu-tion with sorbitol, prior to MALDI measurement the glycatedpeptides needed to be desalted using different sorbents.

Desalting of glycated peptides eluted from tips with sorbitol

For the desalting of glycated peptides eluted from boronateaffinity tips, three different solid phase extraction (SPE) sorbentswere tested. Here 5 mg from fullerene-derivatized silica particlesmade from a silica with a pore diameter of 30 nm (C60(30) silica)and octadecyl silica (C18 silica) was packed in SPE cartridges (All-tech Extract–Clean SPE 1.5-ml reservoir, Alltech Associates, Deer-field, IL, USA). The cartridges were first washed with 2 ml oftetrahydrofuran and then with 1 ml of methanol to eliminate thepossibility of contaminants. Activation of the cartridges was car-ried out with 2 ml of ACN containing 0.1% TFA. Furthermore, theparticles were washed with 2 ml of 50% (v/v) ACN and 0.1% TFAin water and were equilibrated with 2 ml of 0.1% TFA in water. Elu-tion of the bound peptides was implemented with 300 ll of 80%ACN and 0.1% TFA in water. The eluate was evaporated to dryness,and the peptides were taken up in 5 ll of 0.1% TFA in double dis-tilled water. Then 1 ll of this solution was deposited on a stainlesssteel target mixed with 1 ll of matrix.

The desalting of glycated peptides was also carried out usingthe commercially available ZipTip (Millipore). For the activation,equilibration, and elution, the same conditions were used as forthe C60(30) and C18 silica materials. In the case of ZipTip, the elu-ate was deposited directly onto a stainless steel target in 1 ll of80% ACN and 0.1% TFA in water and then mixed with 1 ll of matri-ces and analyzed.

Preparation of different matrices

Two different matrices were compared for the ionization of gly-cated peptides. DHB was prepared in 20% (v/v) ACN and 0.1% TFA inwater at a concentration of 25 mg/ml.

A mixed matrix consisting of DHB and CHCA was also recom-mended for the MALDI–time-of-flight (TOF) analysis of glycatedpeptides [27]. CHCA (2 mg) was dissolved in 100 ll of 70% ACNand 5% formic acid. In addition, 2 mg of DHB was dissolved in100 ll of 70% ACN and 0.1% TFA, and these two solutions weremixed in a 1:1 ratio.

For the measurement of glycated proteins, a saturated SA ma-trix was prepared in 50% (v/v) ACN and 0.1% TFA in water.

Conditions of electrochemical measurements

When using a wall-jet-type detector cell in flow injection anal-ysis (FIA) mode, the electrochemical measurements were made in100 mM sodium hydroxide solutions at a flow rate of 2 ml min�1.In the case of using an enzyme sensor, the glucose measurementwas carried out in a 10 ml electrochemical cell applying a three-electrode cell ensemble. Then 5 ml of isotonic PBS buffer (pH 7.4)was pipetted into the cell. During the measurement, 0.65 V ofworking potential was used while the electrolyte solution was stir-red continually.

MALDI–TOF/MS conditions

All mass spectra were monitored in positive mode with pulsedionization (k = 337 nm, nitrogen laser, maximum pulserate = 50 Hz, maximum intensity = 20–30% of laser for peptides).Peptides of the digests were measured in reflectron mode using a


delayed extraction of 120 ns, and proteins were measured in linearmode at a delayed extraction of 550 ns. The accelerating voltagewas set to +19 kV, and the reflectron voltage was set to +20 kV.Spectra of peptides and proteins were the sum of 1000 shots, andexternal calibration was implemented. Post-source decay experi-ments were also carried out for all peptides with mass shifts ofapproximately 162 and 324 Da to confirm the modifications. Bothfor the parent ions and fragments, 1000 shots were collected. Dataprocessing was executed with Flex Analysis software (version 2.4).For the in silico digestion, Sequence Editor software (Bruker Dal-tonics) was used with the following criteria: (i) all cysteines weresupposed to be treated with iodoacetamide; (ii) monoisotopicmasses were allowed; and (iii) the maximum number of missedcleavage sites was two.

l-LC–MS conditions

A poly(p-methylstyrene-co-1,2-bis(p-vinylphenyl)ethane)-based stationary phase [28] was used for the separation of the pep-tides of digests. The separation was performed under reversedphase conditions with eluent A (0.1% formic acid in water) and elu-ent B (0.1% formic acid in 30% ACN) at a flow rate of 1 ll min�1 and40 �C. A linear gradient (0–50 min: 0% B ? 60% B) was used. Massspectrometric measurements were performed in positive mode.The source voltage was adjusted at 1.4 kV, the capillary tempera-ture was kept at 220 �C, the capillary voltage was at 37 V, andthe tube lens was set at 94 V. When analyzing the results with Se-quest, the following parameters were adjusted: carbamidomethy-lation on C as fixed modification, 144 and 162 (fragments ofglucose) variable modifications at K and R, and three possiblemissed cleavage sites were allowed.

Results and discussion

Evaluation of binding conditions of boronate affinity tips

Electrochemical measurements proved to be a useful approachto optimize the ionic strength and pH of the binding buffer as wellas the most appropriate circumstances of elutions. Fig. 1A displaysthe effect of the pH of binding buffer on the amount of eluted glu-cose. These results clearly demonstrate that the affinity tips bindthe highest amount of glucose when the pH of the binding buffer

Fig. 1. (A) Relationship between the bound glucose and the pH of the binding buffer. (B) A(C) Elution of glucose by formic acid at different pH values. (D) Elution of glucose with

varies between 7.7 and 8.2. Further enhancement of the pH ofthe binding buffer results in a decrease of the bound glucose. Themaximal binding capacity of the tips was found to fall within thepH range of 7.8–8.2 in the curve; therefore, a value of approxi-mately 8.2 was selected for this study. As demonstrated byFig. 1B, when the binding capacity of the tips was plotted againstthe ionic strength of the binding buffer, 150 mM ammonium chlo-ride provided the best results. These results can be regarded asconsistent with data described previously [16,17]; however, thepreviously reported values were not yet based on quantitative evi-dence. Fig. 1C shows the effect of the pH of the eluent depicted inthe function of the eluted glucose when using solutions of formicacid at different pH values for the elution. This relation is exponen-tial; the lower the pH of the eluent, the higher the amount of glu-cose eluted. The elution of bound glucose on boronate-derivatizedresin can also be accomplished by means of a highly concentratedsorbitol solution. The effects of the concentrations of sorbitol solu-tion are plotted against the amount of released glucose in Fig. 1D.The curve reaches a plateau at 1.2 M; therefore, there is no need toemploy sorbitol solutions for the elution at higher concentrations.

Evaluation of performance of boronate affinity tips toward glycatedpeptides enriched from glycated RNase A and HSA tryptic digests

The application of the boronate affinity tips was introducedemploying in vitro glycated HSA and RNase A as model proteins.Fig. 2B shows the effect of the incubation time with glucose onthe degree of the modification of proteins. This can be calculatedfrom the mass shift measured between the unmodified and glycat-ed proteins. Taking into account the fact that the condensation of 1glucose unit can cause approximately a 162 Da mass increase incomparison with the unmodified proteins (Fig. 2A), glycation ofRNase A for 14 days clearly represents that, on the average, 3 glu-cose units were condensed on an RNase A molecule.

HSA was incubated with a high concentration of glucose in PBSbuffer for 12 and 28 days. The reason why the shorter incubationtime was also chosen is that in patients being poorly controlledand suffering from type 2 diabetes mellitus the HSA is also glycat-ed, but the number of glucose moieties condensed on an HSA mol-ecule is not consistent with an overglycated HSA (incubated for28 days). In contrast to Fig. 2C, where the mass spectrum of thenonglycated HSA is shown, Fig. 2D indicates a mass shift of

mount of bound glucose plotted as a function of the ionic strength of binding buffer.sorbitol solution at different concentrations.

Fig. 2. Effect of glycation on molecular weight of proteins. (A) Mass spectrum of unglycated RNase A. (B) Mass spectrum of RNase A glycated for 14 days. (C) Mass spectrum ofunglycated HSA. (D) Mass spectrum of HSA glycated for 12 days. (E) Mass spectrum of HSA glycated for 28 days. Each spectrum was acquired in linear mode and a sum of1000 shots. The applied matrix was SA. a.u., arbitrary units.


1050 Da. This means that, on the average, 7 or 8 glucose moietiesare attached to an HSA molecule. Therefore, HSA glycated for12 days can be considered as a model of HSA being glycated dueto the elevated blood glucose level and isolated from patients un-der poor diabetic control [29]. Fig. 2E represents a mass spectrumbelonging to the overglycated HSA. The corresponding mass shiftshows that approximately 31 glucose moieties are attached to anHSA molecule. This result is relevant to the results reported in pre-vious studies [7,29].

Evaluation of the results measured for the glycated peptideswas carried out using software developed and successfully em-ployed first by the authors (http://www.fraki.lgx.hu) [30]. Thecomparison of two series of masses—the in silico generated andthe measured—can be carried out within a mass tolerance rangeof 50–150 ppm. In this study, 100 ppm mass tolerance was permit-ted. If the measured values provided higher margins comparedwith in silico masses, they were not included in the group of iden-tified peptides. Further identification of glycated constituents oc-curred in their post-source decay spectra. Post-source decayspectra of the glycated peptides yield a specific fragmentation pat-tern so long as neutral losses of a dehydrated glucose (M+H+–162)and a fragment of a glucose (M+H+–120) serve as a basis for theidentification of the modification. From these two neutral losses,the structure of C4H8O4 can be assigned to the latter. Moreover,the series of y ions appearing in the low mass range furnishesimportant information about the sequence of the peptide of inter-est (see Figs. 9 and 10 in Supplementary material).

Although the mixture of two matrices (CHCA and DHB) wasproposed to be used for the analysis of peptides modified with sug-ars, for the glycated peptides DHB matrix at a concentration of25 mg/ml was proven to be better in terms of the number of mon-itored peptides and the sites of modifications localized in them. Forinstance, when the enrichment of a 20 pmol digest of HSA glycatedfor 28 days was carried out using ammonium chloride/ammonia as

binding buffer and the eluate was desalted on C60(30) particlesafter the elution of glycated peptide, the application of DHB re-sulted in 41 single glycated peptides, contrary to the mixture ofDHB and CHCA, where only 30 were found. This comparison wasmeasured for each experiment conducted throughout this study,and in all cases DHB was considered as a more suitable matrix thanthe mixture.

Fig. 3 demonstrates the workflow accomplished throughout thisstudy. As be seen, glycated authentic proteins underwent trypticdigestion. The optimized conditions, including the ionic strengthand pH of the binding buffer and the ways of the different typesof elutions, were applied for HSA glycated for 28 days. This iswhy the highest number of glycated peptides is expected in thiscase; therefore, the differences among the applied approaches ta-ken are obviously better expressed. Results received from theexperiments carried out with a sorbitol solution at a concentrationof 1.2 M were evaluated first. Under optimized conditions, thebinding of the glycated peptides was carried out using ammoniumchloride/ammonia buffer and with taurine buffer using both at thesame pH value (8.2) and the same concentration (150 mmol). Afterthe elution of glycated peptides with sorbitol prior to MALDI–TOFanalysis, the eluate consequently needed to be desalted. This wasimplemented using three different types of sorbents: the commer-cially available ZipTip, fullerene(C60)-derivatized silica material(abbreviated as C60(30) [31] and made on the basis of silica witha 30-nm pore radius), and a homemade densely coated C18 silica[32]. The two latter types were evaluated in detail in SPE experi-ments carried out to separate tryptic peptides and their glycatedderivatives in a previous work by the authors. Results of that workprovided a line of evidence for the excellent binding affinity ofC60(30) toward constituents with enhanced hydrophilic propertiessuch as glycated peptides [30]. The experimental circumstances,including the composition of the binding buffer, the comparisonof the elution using sorbitol and formic acid, and the different

http://www.fraki.extra.hu


types of sorbents employed for desalting the eluates after using asorbitol solution at a concentration of 1.2 M, were evaluated onthe basis of the identified single and double glycated peptidesand the number of possible glycated sites on HSA glycated for28 days. These can be plotted by means of Venn diagrams.Fig. 4A and B depicts the numbers of single glycated peptides cap-tured by the boronate affinity tips employing ammonium chloride/ammonia buffer and taurine buffer, respectively, for the enrich-

Fig. 3. Workflow of different appr

Fig. 4. Comparison of different sorbents used for desalting when elution is carried outpeptides when the loading buffer was ammonium chloride/ammonia at optimized cirpeptides in the case of using taurine buffer. (C) Venn diagram of possible glycated sitesammonia buffer. (D) Venn diagram of possible glycated sites monitored from single gly

ment of glycated constituents. In both cases, the bound glycatedpeptides were eluted with a highly concentrated sorbitol solution(1.2 M) and desalted on the above-mentioned three materials. Asshown in the number of glycated peptides bound and eluted fromeach sorbent, ammonium chloride/ammonia buffer proved to bemore efficient than taurine buffer in that 67 single glycated pep-tides could be selectively captured by means of using ammoniumchloride/ammonia buffer, whereas 55 peptides modified with only

oaches throughout this work.

with sorbitol. (A) Venn diagram showing the numbers of detected single glycatedcumstances. (B) Venn diagram depicting the numbers of detected single glycatedmonitored from single glycated peptides bound by means of ammonium chloride/

cated peptides bound using taurine buffer.


1 sugar unit were possible to be monitored after the elution fromall of the employed phases. The evaluation of binding bufferswas also implemented for glycated peptides modified at two ami-no acid residues with D-glucose (double glycated peptides). The re-sults demonstrated in Fig. 5A and B confirm the observation madein the case of single glycated peptides. It was possible to bind 42double glycated peptides selectively by applying an ammoniumchloride/ammonia buffer. Considerably fewer (33) double glycatedpeptides were successfully monitored in the case of taurine buffer.Not only do the numbers of bound modified constituents indicatethe effect of the binding buffers on the efficiency of the enrich-ment, but also the numbers of possible modification sites locatedin the bound peptides must be taken into account by the evalua-tion. In single glycated peptides, 78 possible sites of modificationwere recognized using ammonium chloride/ammonia buffer(Fig. 4C) and 68 glycated residues were identified by means of tau-rine buffer (Fig. 4D). Regarding the glycated sites identified, in dou-ble modified peptides only a slight difference is observed in thenumber of glycated residues when comparing the two loading buf-fers. The use of the binding buffer consisting of ammonium chlo-ride and ammonia allowed us to identify 63 potential glycatedsites from those peptides captured by boronate affinity tips(Fig. 5C). In comparison, taurine buffer is capable of assisting withbinding peptides to affinity tips, from which 60 glycated sites couldhave been determined (Fig. 5D). In conclusion, the choice of anammonium chloride/ammonia buffer with optimized pH and ionicstrength enables binding more glycated peptides than the exten-sively used taurine buffer with the same pH and concentration.

It is important to emphasize that the differences of using thesebuffer systems are not expressed exclusively in the number ofbound peptides and the identified glycated sites. For instance, afterthe binding of glycated peptides by applying both buffer systems,the elution of the bound constituents was accomplished with a sor-bitol solution. This was followed by a desalting step using SPE car-tridges filled with C60(30) silica particles. After the desalting step,13 unique peptides of the total single glycated peptide pool ad-

Fig. 5. Comparison of different sorbents used for desalting when elution is carried outpeptides when the loading buffer was ammonium chloride/ammonia under optimizedpeptides detected in the case of using taurine buffer. (C) Venn diagram of the possible gchloride/ammonia buffer. (D) Venn diagram of the possible glycated sites detected in do

sorbed by the C60(30) particles (from 41 peptides) were possibleto acquire using ammonium chloride/ammonia as binding buffer(Fig. 4A). Fig. 4B shows unique single glycated peptides also boundexclusively by C60(30) particles when the enrichment of the gly-cated peptides was carried out using taurine buffer. In total, 43of the eluted peptides could be identified when desalting themby means of employing C60(30) silica material. Of 43 single glycat-ed peptides, 15 were identified only from the eluate of C60(30) andthe remaining 28 peptides of the other sorbents could also be de-tected. It may be important to emphasize that these 15 unique gly-cated peptides must be interpreted only in the context ofevaluating the three different sorbents with respect to the numberof bound single glycated peptides. If the unique peptides detectedin the eluates of C60(30) SPE material after the enrichment withthe two above-mentioned buffer systems are compared, it can beconcluded that their numbers were very similar to each other solong as 13 of them were found when enriching them with anammonium chloride/ammonia buffer and 15 glycated peptideswere found when the binding buffer was composed of taurine.Only 5 single glycated peptides with the same sequence and mod-ifications could be analyzed from the unique peptides enrichedwith the two buffers and desorbed from C60(30). The majority ofthe unique peptides were found to be different despite the fact thatthey had been eluted from the same sorbent. This observation re-veals that the use of different buffer systems alters the selectivityof the boronate phase insofar as the captured peptide profile andthe identified potential glycated sites show great variety.

Tables 1 and 2 provide detailed information about the uniqueglycated peptides and the locations of the recognized possible gly-cated sites of these peptides. Using ammonium chloride/ammoniabuffer for the enrichment of glycated constituents with boronateaffinity tips, as described earlier, 41 single and 24 double glycatedpeptides were bound by C60(30) silica particles. In addition, 13 sin-gle and 13 double glycated unique peptides were explored in theeluted peptide pool. Single glycated peptides involve 8 possibleglycated sites, namely K106, R114, R257, R348, K372, K436, K439, and

with sorbitol. (A) Venn diagram showing the numbers of detected double glycatedcircumstances. (B) Venn diagram demonstrating the numbers of double glycatedlycated sites detected in double glycated peptides bound by means of ammoniumuble glycated peptides bound employing taurine buffer.

Table 1Detected single and double unique glycated peptides in the case of using ammoniumchloride/ammonia binding buffer.

Location C60(30) ZipTip C18 silica

(A) Single glycated peptidesUnique peptides 13 12 9Uunique sites 8 4 5

[5–20] R10 or K12 or K20 +[82–98] K93 or R98 +[99–114] K106 or R114 +[115–136] R117 or K136 +[137–145] K137 or R144 or R145 +[161–181] K162 or K174 or K181 +[175–181] K181 +[198–209] K199 or K205 or R209 +[200–212] K205 or R209 or K212 +[206–212] R209 or K212 +[223–233] K225 or K233 +[226–233] K233 +[241–262] R257 or K262 +[275–286] K276 or K281 or K286 +[314–323] K317 or K323 +[318–336] K323 or R336 +[324–336] R336 +[338–351] R348 or K351 +[352–372] K359 or K372 +[360–389] K372 or K378 or K389 +[390–410] K402 or R410 +[403–414] R410 or K413 or K414 +[411–428] K413 or K414 or R428 +[415–432] R428 or K432 +[429–439] K432 or K436 or K439 +[446–466] K466 +[476–484] R484 +[501–524] K519 or R521 or K524 +[520–525] K524 or K525 +[526–541] K534 or K536 or K538 or K541 +[535–545] K536 or K538 or K541 or K545 +[539–557] K541 or R545 or K557 +[546–564] K557 or K560 or K564 +[565–585] K573 or K574 +

(B) Double glycated peptidesUnique peptides 13 11 6Unique sites 7 13 4

[94–114] R98 or K106 or R114 +[99–136] K106 or R114 or R117 or K136 +[138–145] R144 and R145 +[161–181] K162 or K174 or K181 +[163–181] K174 and K181 +[182–190] R186 and K190 +[187–195] K190 and K195 +[213–225] R218 or R222 or K225 +[234–262] K240 or R257 or K262 +[258–276] K262 or K274 or K276 +[263–286] K274 or K276 or K281 or K286 +[275–286] K276 or K281 or K286 +[277–286] K281 and K286 +[318–337] K323 or R336 or R337 +[360–389] K372 or K378 or K389 +[373–402] K378 or K389 or K402 +[411–428] K413 or K414 or R428 +[429–436] K432 and K436 +[429–439] K432 or K436 or K439 +[433–444] K436 or K439 or K444 +[437–444] K439 and K444 +[467–475] R472 and K475 +[476–500] R484 or R485 or K500 +[485–521] R485 or K500 or K519 or R521 +[501–524] K519 or R521 or K524 +[520–525] K524 and K525 +[522–534] K524 or K525 or K534 +[542–560] R545 or K557 or K560 +[546–560] K557 and K560 +

Note. Peptides were released from tips using sorbitol and desalted on C60(30),ZipTip, and homemade C18 silica. Italic bold type indicates unique glycation sites.

Table 2Detected single and double unique glycated peptides in the case of using taurinebinding buffer.

Location C60(30) ZipTip C18 silica

(A) Single glycated peptidesUnique peptides 15 8 4Unique sites 9 6 0

[115–136] R117 or K136 +[115–137]a R117 or K136 or K137 +[115–144] R117 or K136 or K137 or R144 +[161–181] K162 or K174 or K181 +[163–174] K174 +[163–181] K174 or K181

[175–181] K181 +[175–186] K181 or R186

[175–190] K181 or R186 or K190 +[200–212] K205 or R209 or K212 +[206–212] R209 or K212 +[226–257] K233 or K240 or R257 +[241–262] R257 or K262 +[241–274] R257 or K262 or K274 +[258–276] K262 or K274 or K276 +[275–286] K276 or K281 or K286 +[287–323] K313 or K317 or K323 +[314–323] K317 or K323 +[352–372] K359 or K372 +[390–402] K402

[390–410] K402 or R410 +[403–414] R410 or K413 or K414 +[411–428] K413 or K414 or R428 +[415–432] R428 or K432 +[440–445] K444 or R445 +[446–466] K466 +[485–519] R485 or K500 or K519 +[526–541] K534 or K536 or K538 or K541 +[539–557] K541 or R545 or K557 +[542–560] R545 or K557 or K560 +

(B) Double glycated peptidesUnique peptides 5 7 7Unique sites 8 12 8

[5–12] R10 and K12 +[74–98] R81 or K93 or R98 +[94–114] R98 or K106 or R114 +[99–136] K106 or R114 or R117 or K136 +[115–136] R117 and K136 +[145–160] R145 or K159 or R160 +[161–181] K162 or K174 or K181 +[163–181] K174 and K181 +[213–222] R218 and R222 +[223–240] K225 or K233 or K240 +[263–286] K274 or K276 or K281 or K286 +[360–389] K372 or K378 or K389 +[414–428] K414 and R428 +[437–444] K439 and K444 +[445–466] R445 and K466 +[526–538] K534 or K536 or K538 or K541 +[539–557] K541 or R545 or K557 +[561–573] K564 and K573 +[561–574] K564 or K573 or K574 +

Note. Peptides were released from tips using sorbitol and desalted on C60(30),ZipTip, and homemade C18 silica. Italic bold type indicates unique glycation sites.

a Peptide in the sequence bracket [115–137] could be detected from the digest ofnonglycated HSA (standard).


K466, whereas 7 unique sites modified with D-glucose were identi-fied from double glycated peptides, namely R218, R484, R485, K500,K519, R521, and K560 (Figs. 4 and 5 and Table 1).

These results were compared with those received after theenrichment of glycated peptides using taurine buffer. Among thosepeptides released from the affinity tips and desalted from the sor-bitol solution on C60(30) particles, 15 single and 5 double glycatedunique peptides were analyzed. Single glycated peptides madepossible the identification of 9 unique glycated sites. Surprisingly,


of the 5 double glycated unique peptides, 8 unique possible mod-ified sites were recognized, as shown in Figs. 4 and 5 and Table2. In single glycated unique peptides K190, K372, K413, K444, R445,K466, K536, K538, and K541, unique glycated sites were localized. Indouble glycated unique peptides, residues at R81, K93, R218, K372,K378, K389, R445, and K466 were likely to undergo nonenzymaticglycation.

ZipTip is a commercially available product, probably the mostfrequently and efficiently used tool for desalting the solutions ofproteins and peptides prior to mass spectrometric measurements.In previous studies, the superiority of C60(30) to other commercialproducts such as Oasis and Sep-Pak and some octadecyl and tria-conthyl modified silica material with high surface coverage hasbeen reported in terms of SPE experiments of different biomole-cules, including peptides, proteins, phosphorylated peptides, andglycated peptides [31]. Through this work, an idea of trying to firstuse ZipTip for desalting glycated peptides emerged. When bindingpeptides to affinity tips with ammonium chloride/ammonia buffer,the desalting procedure carried out with ZipTip allowed the iden-tification of 12 single glycated unique peptides modified at the res-idues of K93, R98, K402, and K524. Thus, in terms of the number ofunique glycated sites localized in single unique glycated peptides,ZipTip does not seem to be as efficient as C60(30); however, itmust be noted that 42 single glycated peptides were desalted bymeans of ZipTip, and 58 possible glycated sites were monitored.In addition, the analysis of 11 double glycated unique peptidescomprising 13 glycated unique sites was made possible by ZipTip.These peptides were modified at the residues of R144, R145, K162,R186, R257, K262, K323, K372, K378, K389, K402, R472, and K475.

Concerning taurine binding buffer, for single glycated peptides,ZipTip provided a worse result than C60(30) so long as only 30peptides were bound; of these, 8 were found to be unique. Takinginto account the glycated residues detected in peptides bound onZipTip, 54 possible glycated sites were found, in contrast toC60(30), where this number was slightly higher (61). Six uniqueglycated sites at the positions of K402, K432, K313, K317, K500, and

Fig. 6. Comparison of elutions of glycated peptides from affinity tips with digest of HSA gdiagram showing the numbers of single glycated peptides using ammonium chloride/ammthe tips in the case of employing taurine binding buffer. (C) Venn diagram of the doublVenn diagram depicting double glycated peptides comparing the elutions with the dige

R485 could be observed. Regarding the number of double glycatedpeptides, ZipTip proved to be as efficient as C60(30) insofar as bothsorbents were capable of binding 16 peptides. For ZipTip, 7 uniquedouble glycated peptides with 12 unique sites at K106, R114, R145,K159, R160, K414, R428, K439, K444, K534, K536, and K538 were detected.

Finally, the desalting of the glycated peptides was also triedwith octadecyl silica having high surface coverage. In the case ofammonium chloride/ammonia buffer 36 single glycated peptideswere successfully bound, 9 of which were unique peptides, modi-fied at K199, K317, K276, K564, and K573 as well. Using this buffer, only13 double glycated peptides could be desalted. The poor perfor-mance of this material toward hydrophilic glycated peptides is alsoexpressed in the number of unique peptides found, that is, only 6with 4 possible unique glycated sites at R98, K413, K414, and R428.By means of taurine buffer, the results of the desalting on C18 par-ticles were even worse. In total, 28 single glycated peptides werebound; of the 4 unique peptides analyzed, none of them comprisedany unique glycated sites. In total, 17 double glycated peptidescould be identified; of these, 7 proved to be unique with 8 possiblesites of glycation at R10, K12, K162, K274, K276, K564, K573, and K574.

Elution of the bound constituents is frequently implemented inacidic media. As reported earlier, a formic acid solution of pH 2.0proved to be the best in terms of the amount of eluted D-glucoseevidenced by amperometric measurements. Elution with this solu-tion was compared with elution implemented with sorbitol (de-salted by C60(30)) regarding the number of eluted single anddouble glycated peptides and the possible identified glycated sitesas well. These two ways of elution were compared with the digestof HSA incubated with D-glucose for 28 days to assess the efficiencyof the enrichment. Surprisingly, the elution with formic acid pro-vided the worst results in terms of the number of identified glycat-ed peptides and the possible sites of glycations.

Fig. 6A shows that in 100 pmol of the digest of overglycated HSAwithout enrichment, 29 single glycated peptides were recognizedand 3 were found to be unique peptides with 5 unique glycatedsites. When binding glycated peptides with ammonium chloride/

lycated for 28 days in terms of the numbers of detected glycated peptides. (A) Vennonia binding buffer. (B) Venn diagram of the single glycated peptides released from

e glycated peptides detected when using ammonium chloride/ammonia buffer. (D)st when the binding buffer was taurine.


ammonia buffer, the elution of single glycated peptides with for-mic acid resulted in the release of 27 single glycated peptides fromthe boronate affinity tips; of these, 7 were unique peptides with 4possible sites of modification. As mentioned earlier, the elutionwith sorbitol yielded the detection of 41 single glycated peptides;nevertheless, the eluate was subjected to SPE on C60(30) particlesprior to MALDI–TOF analysis. However, in this context of the com-parison, 15 unique peptides comprising 12 unique residues wereexplored. Using the same buffer, the same trend could be observedfor double glycated peptides given that 14 of them were identifiedfrom the digest (only 3 of them were unique with 3 unique sites).As also demonstrated in Fig. 6B, formic acid made possible the re-lease of only 11 peptides. Of these double glycated peptides, 3 werefound to be unique, including 5 unique sites of glycation. Elutionwith sorbitol again showed the best results after desalting. Thiswas expressed by 24 identified peptides; of these, 14 were ob-served to be unique constituents containing 16 unique glycatedsites.

Fig. 6C and D show those cases where glycated peptides werebound to the affinity tips in the presence of taurine buffer. Briefly,from the digest, 29 single glycated peptides were detected; ofthese, 6 were unique and had 3 unique residues. Eluting the pep-tides with formic acid solution allowed the recognition of 31 singleglycated peptides; of these, 8 were unique with 4 possible uniquesites of glycation. By means of eluting the peptides with sorbitol,43 single glycated peptides could successfully be detected; ofthese, 15 were unique and incorporated 8 unique glycated sites.With respect to double glycated peptides in the case of using tau-rine buffer, 14 were found in the digest; of these, 8 were unique

Fig. 7. Enrichment of glycated peptides. (A) Mass spectrum of digest of HSA glycated for12 days after selective enrichment. (C) Mass spectrum of digest of HSA glycated for 28 dammonium chloride/ammonia binding buffer and the elution of the bound peptides wasC60(30) particles. Each spectrum was monitored in reflectron mode and a sum of 1000arbitrary units.

with 8 unique residues. Whereas elution with formic acid provided9 double glycated peptides (4 of these were unique with 7 uniquesites), elution accomplished with sorbitol caused the release of aconsiderably higher number of peptides so long as 16 could bemeasured after desalting (10 of these proved to be unique andcomprised peptides of 13 unique glycated sites). As demonstratedabove, with the elution of the bound glycated peptides with formicacid solution, it seems as though no enrichment of the modifiedconstituents occurred in spite of the fact that the highest amountof glucose could be measured at pH 2.0 (as discussed earlier).Therefore, the acidic elution is not recommended by the authors.

The most efficient way to gain information about the modifiedglycated sites of a protein by MALDI–MS is to bind glycated pep-tides from the digest of the protein using, for instance, ammoniumchloride/ammonia binding buffer under optimized circumstances,and then the selectively bound constituents are eluted with a con-centrated solution of a proper sugar such as sorbitol. Furthermore,prior to analysis, the presence of undesirable sugar can be elimi-nated using an effective sorbent for desalting. As was demon-strated, C60(30) proved to be as efficient as ZipTip; no significantdifferences could be observed in the binding efficacy of eitherphase toward glycated peptides.

Although the above-mentioned methods were evaluated on thebasis of the number of unique peptides and glycated sites, some ofthese modifications could be present in the nonglycated HSA stan-dard because it was isolated from human serum. Therefore, if theaim of the work is also to identify those possible glycated sitesbeing modified only after the in vitro glycation, the analysis ofthe digest of nonmodified HSA is also an important requirement.

28 days. No enrichment was used. (B) Mass spectrum of digest of HSA glycated forays after selective enrichment. In both cases, the enrichment was carried out usingcarried out with sorbitol. Prior to MALDI measurement, the eluates were desalted onshots. The applied matrix was DHB. Asterisks (*) denote the glycated peptides. a.u.,

Table 3Glycated residues identified by LC–MS.

Location Glycation site

[42–64] K51 or K64

[160–174] R160 or K162 or K174 +[226–240]a K233 or K240

[258–274] K262 or K274

[263–276] K274 or K276

[318–336] K323 or R336

[349–359] K351 or K359

[373–389]a K378 or K389

[390–402] K402

[414–428]a K414 or R428

[473–484] K475 or R484

[525–534]a K525 or K534

[539–557] K541 or R545 or K557 *

[542–557]a K545 or K557

Note. The location assigned by the plus sign (+) means that in this region of HSAboth single and double glycated peptides were detected. The asterisk (*) denotes adouble glycated peptide.

a Peptide could be detected from the digest of nonglycated HSA.


These glycated peptides present in the digest of nonmodified HSAprobably comprise the most privileged accessible residues; there-fore, their identification—especially after the nonenzymatic glyca-tion (when their amounts in the digest are increased immensely)—is not a challenging task. As such, if these peptides are taken intoaccount, the sensitivity of the given method is undermined. Anal-ysis of the digest of nonglycated HSA allowed us to identify 11 pep-tides by means of using ammonium chloride/ammonia as a bindingbuffer; this was followed by an elution with sorbitol, and the eluatewas desalted using both C60(30) and ZipTip (see Table 7 in Supple-mentary material). As expected, elution with formic acid madepossible the identification of only 2 peptides, both of which couldbe analyzed from the eluates with sorbitol. It is also worth notingthat 4 of these peptides contain K233, K378, K525, and K545 glycatedresidues, and these residues are well known to be privileged andare able to be glycated easier [9,26,30]. From the digest of ungly-cated RNase A, no glycated peptides were monitored.

The usefulness of the method proposed by the authors is clearlydemonstrated in Fig. 7. Fig. 7A shows the MALDI spectrum of thedigest of HSA (glycated in vitro for 28 days) in the range from2000 to 2230 Da. Besides a very intensive nonmodified tryptic pep-tide appearing at m/z 2086.46, 2 less intensive glycated peptidescan be observed at m/z 2060.68 and 2013.56. Moreover, the massaccuracy of these peptides is not reliable due to the very low sig-nal-to-noise ratio of the first peptide. The latter does not possessvery good mass resolution, which can be ascribed to the fact thatpeptides with similar masses can have a disturbing effect on massresolution. Fig. 7C shows the spectrum belonging to the digest of

Table 4Comparison of affinity chromatographic–off-line MALDI method with LC–MS in terms of i

Location Glycation site Sequence

[1–10] K1 or K7 or R10 KETAAAKFER[11–31] K31 QHMDSSTSAASSSNYCNQMMK[32–39] R33 or K37 or R39 SRNLTKDR[34–39] K37 or R39 NLTKDR[38–61] R39 or K41 or K61 DRCKPVNTFVHESLADVQAVCSQK[40–61] K41 or K61 CKPVNTFVHESLADVQAVCSQK[62–85] K66 or R85 NVACKNGQTNCYQSYSTMSITDCR[67–91] R85 or K91 NGQTNCYQSYSTMSITDCRETGSSK[92–104] K98 or K104 YPNCAYKTTQANK[99–124] K104 TTQANKHIIVACEGNPYVPVHFDASV

Note. DD refers to the double glycated peptides.

HSA glycated for 28 days after having been enriched by the methodproposed above. Four glycated peptides can be seen in the figure.The peak appearing at m/z 2061.05 is a single glycated peptide.Additional peaks at m/z 2094.06, 2104.03, and 2217.24 are doubleglycated peptides, and their sequences are also shown in Fig. 7C.When the profile of glycated peptides enriched from HSA glycatedfor 12 days (Fig. 7B) is compared with the results of enrichmentfrom the digest of HSA glycated for 28 days (Fig. 7C), significant dif-ferences can be found. For example, as a result of the enrichmentfrom the digest made from HSA glycated for 12 days, 3 double gly-cated peptides were monitored at m/z 2094.28, 2104.09, and2207.45. The rather intensive single glycated peptide detected inthe digest of HSA glycated for 28 days at m/z 2061.05 could notbe detected from the digest of HSA glycated for 12 days. Thismeans that this glycation site must be more hindered; therefore,it needs to be exposed longer to the high concentration of sugar.This observation is also relevant to the peptide detected at m/z2217.25 from the digest of HSA glycated for 28 days. In contrast,the peptide detected at m/z 2207.45 from the digest of HSA glycat-ed for 12 days clearly indicates that this peptide, comprising one ofthe most privileged glycation sites (K378) [26], was modified rela-tively rapidly and that further oxidation processes are responsiblefor the absence of this peptide in the spectrum.

Previously published studies have reported the investigation ofglycated peptides employing LC–tandem mass spectrometry (MS/MS) techniques [10,26]. The advantages of using these approachesin comparison with different off-line MALDI methods are that theloss of samples is largely reduced so long as the separated peptidesare analyzed on-line by mass spectrometric methods. Table 3 indi-cates glycated peptides identified by the authors from the digest ofHSA glycated in vitro for 28 days. To the best of the authors’ knowl-edge, this is the first attempt to separate glycated peptides on amonolithic capillary column. The results are very similar to thosereported in previous studies in terms of the number of glycatedpeptides and the possible modified residues [26]. Table 3 showsthat of the 15 glycated peptides identified, 13 were found to be sin-gle glycated and only 2 were modified with 2 glucose units. How-ever, the table also shows that 5 peptides from all of the identifiedpeptides also were present from the digest of the nonglycated HSA.This fact clearly indicates that despite using a highly efficient sep-aration method, the efficacy of this method is overcome by the bor-onate affinity chromatography/off-line MALDI–TOF methodintroduced in this article.

The peptides identified allowed us to localize 28 possible glyca-tion sites. This proved to be better than previously reported data;of 15 peptides, 22 possible sites of modification were describedby Ref. [26]. However, both LC–MS methods were able to recognizethe privileged K233, K276, K378, K545, and K525 sites of glycation.

dentified glycated peptides of RNase A.

Identified by LC–MS Identified by boronate affinity–MALDI

6 58 9

+ DD+

DD+++ ++

++

+ +


Table 4 shows peptides and glycation sites of RNase A recog-nized by boronate affinity tips (using ammonium chloride/ammo-nia buffer). The bound peptides were eluted with sorbitol anddesalted on C60(30) particles. As can be seen, by means of affinitytips, 5 single and 2 double glycated peptides comprising 12 possi-ble glycated sites could be detected. Even if the focus of the inves-tigation is a small protein such as RNase A where the number ofglycated sites is reletively low, with the help of LC–MS, 6 glycatedpeptides with 8 possible modification sites were measured.

Application of boronate affinity tips for selective enrichment ofAmadori products from digest of HSA collected from sera of patientssuffering from type 2 diabetes mellitus and healthy volunteers

Four healthy volunteers and four patients suffering from type 2diabetes mellitus were chosen to test the method described abovefor the mapping of the level of glycation of HSA. Patients were se-lected on the basis of the level of HbA1c and fasting plasma glu-cose. Because boronate affinity chromatography has beenrecognized to bind constituents comprising cis–diol, the method al-lows monitoring fructosyl lysine (FL, Amadori product) and itsderivative after loss of water (FL-18). For healthy volunteers, be-sides some important clinical parameters, the fasting plasma glu-cose was always less than 6.1 mmol L–1. These values were by farthe highest in the cases of patients 1 and 2 (12.17 and13.02 mmol L–1, respectively). For patients 3 and 4, slightly lowervalues were obtained (8.53 and 7.50 mmol L–1, respectively).

Because HbA1c was used for monitoring long-term glycemiccontrol in patients with diabetes mellitus, this value is thoughtto indicate glycemic state over the most recent 1–2 months. Forall of the investigated patients, HbA1c was considered to be ratherhigh (10.30, 11.50, 11.82, and 11.30% as measured for patients 1, 2,3, and 4, respectively).

These two series of data served for the estimation of the glyce-mic state of patients. HSA reflects the amount of blood glucosemore rapidly than HbA1c; moreover, accounting for its relativelyshort (17 day) biological half-life, the glycation level of HSA is rel-

Table 5Single and double glycated peptides comprised by FL residues (Amadori peptides)detected in patients (P1–P4) and healthy volunteers (H1–H4).

Location Possible glycation site H1 H2 H3 H4 P1 P2 P3 P4

(A) Single glycated peptides[1–12] K4 or R10 or K12 + + + + + +[11–20] K12 or K20 +[52–73] K64 or K73 +[65–81] K73 or R81 + +[137–144] K137 or R144 + + + + + +[191–199] K195 or R197 or K199 + +[219–225] R222 or K225 +[219–233] R222 or K225 or K233 + +[226–240] K233 or K240 + + + + + + + +[263–274] K274 + +[314–323] K317 or K323 + +[318–336] K323 or R336 + + +[337–348] R337 or R348 + + +[373–389] K378 or K389 + + +[390–410] K402 or R410 +[437–445] K439 or K444 or R445 +[525–534] K525 or K534 + + + + + + + +[535–541] K536 or K538 or K541 +[561–573] K564 or K573 +

(B) Double glycated peptides[1–10] K4 and R10 + +[198–209] K199 or K205 or R209 +[277–286] K281 and K286 + + +[403–414] R410 or K413 or K414 +[440–445] K444 and R445 +[525–534] K525 and K534 + + + + + +

evant to the condition of blood glucose over the most recent2 weeks.

Approximately 300 lg of the digest of HSA collected from eachpatient and healthy volunteer was enriched under optimized con-ditions using boronate affinity tips. Also considering the results re-ported earlier, the bound constituents were released from the tipsby means of a 1.2 M sorbitol solution. The eluate was then desaltedon C60(30) and preconcentrated. Peptides remaining after this pro-cedure were measured by MALDI–TOF/MS.

Results reported in Table 5 show single and double glycatedpeptides with the corresponding possible glycation sites beingmonitored in healthy volunteers and patients. Besides the privi-leged glycation sites of K12, K233, and K525 being present in eachindividual, many other glycated K and R residues described previ-ously were recognized [9].

Regarding the numbers of glycated peptides and the possibleidentified glycated residues with the exception of patient 4 (whosefasting plasma glucose was the lowest), the clinical parameters canbe associated with the glycation pattern obtained from mass spec-trometric data.

Table 6 summarizes the explored glycated peptides with a massshift of 144.042. They can be either FL (Amadori product) after aloss of water or tetrahydropyrimidine bound directly to arginineresidues. Both are able to bind to a boronate resin due to the pres-ence of vicinal diol. Results also indicate that peptides appearing atm/z 1077.519 and 1783.938 are worth considering. The latter wasobserved in all of the investigated digests belonging to the patientsbut was utterly absent from samples taken from healthy individu-als. This is clearly presented in Fig. 8. The presence of a modifica-tion at m/z 144.042 was further corroborated by thecorresponding post-source decay spectrum, where the presenceof a neutral loss was detected at m/z 1639.896, showing that thelocation of the modification can be at either K414 or R428. Similarly,peptide at m/z 1077.519 could be detected in only three diabeticpatients. This modification located at R81 is considered to be a tet-rahydropyrimidine attached, as expected, to arginine.

Table 6Peptides modified with glucose after a loss of water and detected in patients (P1–P4)and healthy volunteers (H1–H4).

Location Possible glycation site H1 H2 H3 H4 P1 P2 P3 P4

[65–73] K73 + +[74–81] R81 + + +[137–

145]K137 or R144 or R145 + + +

[187–195]

K190 or K195 +

[191–199]

K195 or R197 or K199 +

[210–218]

K212 or R218

[219–225]

R222 or K225 + + +

[373–389]

K378 or K389 + +

[414–428]

K414 or R428 + + + +

[437–444]

K439 or K444 + +

[501–521]

K519 or R521 +

[520–525]

K524 or K525 + + + + +

[525–538]

K525 or K534 or K536 or K358 +

[535–545]

K536 or K538 or K541 or K545 + +

Note. This modification causes approximately a 144-Da mass shift as compared withthe corresponding tryptic peptides. Italic bold type indicates those glycated sitesthat can have an important role as possible biomarker in type 2 diabetes mellitus.

Fig. 8. Comparison of mass spectra of HSA digest taken from sera of diabetic patients (P1–P4) and healthy volunteers (H1–H4) after enrichment of glycated peptides in therange of 1760–1800 Da. The presence of the peak appearing at m/z 1783.938 in patients suggests the possible distinctive role of this peptide as a biomarker. Each spectrumwas monitored in reflectron mode and a sum of 1000 shots. The applied matrix was DHB. a.u., arbitrary units.


More patients and control individuals need to be involved tomake a decision as to whether these 2 peptides with the possi-ble modifications can serve as potential biomarkers in diabetesmellitus. The method elaborated and reported here seems tobe a powerful tool for the investigation of nonenzymaticglycation.

Conclusions

Microscale boronate affinity tips have proved to be an efficienttool for the selective enrichment of glycated peptides. Data ob-tained from electrochemical measurements allowed us to findthe optimal ionic strength and pH value of the binding buffer.Using an ammonium chloride/ammonia buffer at different pHand ionic strength values for the binding of glucose suggests thatboronate affinity resin binds the highest amount of glucose at pH8.2. The concentration of salt in the buffer was 150 mM, and elu-tion of the bound glucose could be carried out employing eitheracidic conditions or a highly concentrated solution of sorbitol(1.2 M). The latter proved to be better.

The choice of buffer and the way of elution, as well as the mostappropriate sorbent used for desalting the eluate, were evaluatedby means of Venn diagrams when using the digest of HSA glycatedfor 28 days in terms of the numbers of single and double glycatedpeptides as well as the number of possible recognized glycationsites located on these peptides. Surprisingly, the tips showed betterperformance when the binding procedure was carried out in anammonium chloride/ammonia buffer; however, both of the buffersystems used can cause different selectivity of the tips toward gly-cated constituents. This is clearly expressed by the different pro-files of the eluted peptides and glycation sites located in thepeptides. Elution carried out using a sorbitol solution at a concen-

tration of 1.2 M was found to be more efficient than elution withformic acid at pH 2.0. For instance, whereas a solution of formicacid enabled the release of 27 single glycated peptides (of these,7 peptides were recognized to be unique in this context of explana-tion), elution with sorbitol allowed us to identify 41 single glycatedpeptides (of these, 15 were unique). In this study, three differentsorbents—commercial ZipTip, C60(30), and homemade C18 sil-ica—were tried and compared. Regarding the numbers of singleand double glycated peptides and the possible modified residues,C60(30) was considered as a promising sorbent with respect toits performance toward glycated constituents. Moreover, its effi-cacy is competitive with commercial ZipTip. The application ofC18 did not meet the requirements concerning the number of gly-cated peptides bound during the desalting procedure.

Results provided by the LC–MS experiment in the case of ana-lyzing digests of glycated HSA and RNase A were not satisfactoryor comparable to results provided by the affinity chromatographicoff-line MALDI method.

The availability of the method was demonstrated for the digestof HSA isolated from healthy volunteers and diabetic patients.However, both the control and diabetic groups included only fourindividuals, and more glycated peptides, and consequently moreglycated sites, were recognized for patients suffering from type 2diabetes mellitus and having been under insufficient diabetic con-trol when monitoring FL (Amadori product). When detecting FLafter a loss of water (FL-18), a distinctive peptide appearing at m/z 1783.9 was observed in each diabetic patient but was not de-tected in healthy control individuals. This peptide is located inthe region of [414–428] of HSA and incorporates K414 and R428. Thispeptide and its glycated sites may serve as a potential biomarkerfor diabetes; however, additional measurements are required toconfirm this observation.


Acknowledgments

This work was supported by Grants PD 76395 and PD 78599 ofthe Hungarian National Science Foundation (OTKA). The authorsthank Gábor Rébék-Nagy for checking the manuscript carefully.The procedures used for human sera were approved by The EthicsCommittee of the Medical Faculty of the University of Pécs.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, inthe online version, at doi:10.1016/j.ab.2009.06.007.

References

[1] J.W. Baynes, N.G. Watkins, C.I. Fisher, C.J. Hull, S.J. Patrick, M.U. Ahmed, J.A.Dunn, S.R. Thorpe, The Amadori product on protein: structure and reaction,Prog. Clin. Biol. Res. 304 (1989) 43–47.

[2] N. Ahmed, P.J. Thornalley, Quantitative screening of protein biomarkers ofearly glycation, advanced glycation, oxidation, and nitrosation in cellular andextracellular proteins by tandem mass spectrometry multiple reactionmonitoring, Biochem. Soc. Trans. 31 (2003) 1417–1422.

[3] D.R. McCance, D.G. Dyer, J.A. Dunn, K.E. Bailie, R.S. Thorpe, J.W. Baynes, T.J.Lyons, Maillard reaction products and their relation to complication in insulin-dependent diabetes mellitus, J. Clin. Invest. 91 (1993) 2470–2478.

[4] L.A. Trivelli, H.M. Ranney, H.-T. Lai, Hemoglobin components in patients withdiabetes mellitus, N. Engl. J. Med. 284 (1971) 353–357.

[5] P. Ulrich, A. Cerami, Protein glycation, diabetes, and aging, Recent Prog. Horm.Res. 56 (2001) 1–21.

[6] R. Kisugi, T. Kouzuma, T. Yamamoto, S. Akizuki, H. Miyamoto, Y. Someya, J.Yokoyama, I. Abe, N. Hiraki, A. Ohnishi, Structural and glycation site changes ofalbumin in diabetic patient with very high glycated albumin, Clin. Chim. Acta382 (2007) 59–64.

[7] F.L. Brancia, J.Z. Bereszczak, A. Lapolla, D. Fedele, L. Baccarin, R. Seraglia, P.Traldi, Comprehensive analysis of glycated human serum albumin trypticpeptides by off-line liquid chromatography followed by MALDI analysis on atime-of-flight/curved field reflectron tandem mass spectrometer, J. MassSpectrom. 41 (2006) 1179–1185.

[8] C. Wa, R. Cerny, D.S. Hage, Obtaining high sequence coverage in matrix-assisted laser desorption time-of-flight mass spectrometry for studies ofprotein modification: analysis of human serum albumin as a model, Anal.Biochem. 349 (2006) 229–241.

[9] C. Wa, R.L. Cerny, W.A. Clarke, D.S. Hage, Characterisation of glycation adductson human serum albumin by matrix-assisted laser desorption/ionization time-of-flight mass spectrometry, Clin. Chim. Acta 385 (2007) 48–60.

[10] H.S. Gadgil, P.V. Bondarenko, M.J. Treuheit, D. Ren, Screening and sequencingof glycated proteins by neutral loss scan LC/MS/MS method, Anal. Chem. 79(2007) 5991–5999.

[11] P.M. Danze, A. Tarjoman, J. Rousseaux, P. Fossati, M. Dautrevaux, Evidence foran increased glycation of IgG in diabetic patients, Clin. Chim. Acta 166 (1987)143–153.

[12] J. Rohovec, T. Maschmeyer, S. Aime, J.A. Peters, The structure of the sugarresidue in glycated serum albumin and its molecular recognition byphenylboronate, Chem. Eur. J. 9 (2003) 2193–2199.

[13] X.-C. Liu, Boronic acids as ligands for affinity chromatography, Chin. J.Chromatogr. 24 (2006) 73–80.

[14] J. Heon, Y. Kim, M.Y. Ha, E.K. Lee, J. Choo, Immobilization ofaminophenylboronic acid on magnetic beads for the direct determination of

glycoproteins by matrix assisted laser desorption ionization massspectrometry, J. Am. Soc. Mass Spectrom. 16 (2006) 1456–1460.

[15] O.G. Potter, M.C. Breadmore, E.F. Hilder, Boronate functionalised polymermonoliths for microscale affinity chromatography, Analyst 131 (2006) 1094–1096.

[16] F.A. Middle, A. Bannister, A.J. Bellingham, P.D.G. Jean, Separation of glycatedhemoglobins using immobilized phenylboronic acid, Biochem. J. 209 (1983)771–779.

[17] Q. Zhang, N. Tang, J.W.C. Brock, H.M. Mottaz, J.M. Ames, J.W. Baynes, R.D.Smith, T.O. Metz, Enrichment and analysis of nonenzymatically glycatedpeptides: boronate affinity chromatography coupled with electron-transferdissociation mass spectrometry, J. Proteome Res. 6 (2007) 2323–2330.

[18] A. Monzo, G.K. Bonn, A. Guttman, Boronic acid–lectin chromatography: I.Simultaneous glycoprotein binding with selective or combined elution, Anal.Bioanal. Chem. 389 (2007) 2097–2102.

[19] S. Gontarev, V. Shmanai, S.K. Frey, M. Kvach, F.J. Schweigert, Application ofphenylboronic acid modified hydrogel affinity chips for high-throughput massspectrometric analysis of glycated proteins, Rapid Commun. Mass Spectrom.21 (2007) 1–6.

[20] C. Leeuwenburgh, P. Hansen, A. Shaish, J.O. Holloszy, J.W. Heinecke, Markers ofprotein oxidation by hydroxyl radical and reactive nitrogen species in tissuesof aging rats, Am. J. Physiol. 274 (1998) R453–R461.

[21] X. Li, J. Pennington, J.F. Stobaugh, C. Schöneich, Synthesis of sulfonamide– andsulfonyl–phenylboronic acid-modified silica phases for boronate affinitychromatography at physiological pH, Anal. Biochem. 372 (2008) 227–236.

[22] Y.C. Li, J.-O. Jeppson, M. Jörtén-Karlsson, E.L. Larsson, H. Jungvid, I.Y. Galaev, B.Mattiasson, Application of shielding boronate affinity chromatography in thestudy of the glycation pattern of haemoglobin, J. Chromatogr. B 776 (2002)149–160.

[23] X.C. Liu, W.H. Scouten, Boronate affinity chromatography, Methods Mol. Biol.147 (2000) 119–128.

[24] L. Nagy, G. Nagy, P. Hajós, Copper electrode based amperometric detector cellfor sugar and organic acid measurements, Sens. Actuat. B 76 (2001) 493–498.

[25] L. Nagy, N. Kálmán, G. Nagy, Periodically interrupted amperometry: a way ofimproving analytical performance of membrane coated electrodes, J. Biochem.Biophys. Methods 69 (2006) 133–141.

[26] A. Lapolla, D. Fedele, R. Reitano, N.C. Arico, R. Seraglia, P. Traldi, E. Marotta, R.Tonani, Enzymatic digestion and mass spectrometry in the study of advancedglycation end products/peptides, J. Am. Soc. Mass Spectrom. 15 (2004) 496–509.

[27] S. Laugesen, P. Roepstroff, Combination of two matrices results in improvedperformance of MALDI MS for peptide mass mapping and protein analysis, J.Am. Soc. Mass Spectrom. 14 (2003) 992–1002.

[28] L. Trojer, G.K. Bonn, Fabrication of organic monolithic columns for theseparation of biopolymers and small molecules by modulation of thepolymerization time, U.S. Patent, 11/551,181 (2006).

[29] P. Thornalley, M. Argirova, N. Ahmed, V.M. Mann, O. Argirov, A. Dawnay, Massspectrometric monitoring of albumin in uremia, Kidney Int. 58 (2000) 2228–2234.

[30] K. Böddi, A. Takátsy, S. Szabó, L. Markó, I. Wittmann, R. Ohmacht, G. Montskó,R.M. Vallant, T. Ringer, R. Bakry, C.W. Huck, G.K. Bonn, Z. Szabó, Use offullerene, octadecyl, and triaconthyl silica for solid phase extraction of trypticpeptides obtained from unmodified and in vitro glycated human serumalbumin and fibrinogen, J. Sep. Sci. 32 (2009) 295–308.

[31] R.M. Vallant, Z. Szabó, S. Bachmann, R. Bakry, M.N. Ul-Haq, M. Rainer, N. Heigl,C. Petter, C.W. Huck, G.K. Bonn, Development and application of C60–fullerenebound silica for solid phase extraction of biomolecules, Anal. Chem. 79 (2007)8144–8153.

[32] Z. Szabó, R. Ohmacht, C.W. Huck, W.M. Stöggl, G.K. Bonn, Influence of the porestructure on the properties of silica based reversed phase packings for LC, J.Sep. Sci. 28 (2005) 313–324.

http://dx.doi.org/10.1016/j.ab.2009.06.007

Enrichment of Amadori products derived from the nonenzymatic glycation of proteins using microscale...

Documents

Transcript of Enrichment of Amadori products derived from the nonenzymatic glycation of proteins using microscale...